In 1965, Gordon Moore predicted that the number of components in complex integrated circuits would double every two years, for a CAGR of 41%. In particular, he was talking about the number of transistors on microchips. As it turned out, he was right, and his prediction became “Moore’s Law”. It’s become synonymous with a kind of confident technological optimism: we expect computational progress to continue unabated, and for processes to get better, faster, and easier over time.
Cut to the world of drug discovery, where we would really hope to see this taking place, and it just isn’t. Drug discovery is actually becoming harder, slower and more costly over time, so dramatically that researchers have coined “Eroom’s Law”, the inverse of “Moore’s Law” to describe how bad things really are. Proponents of this view estimate that the cost of developing a new drug doubles every nine years. Diminishing returns indeed.
How do we flip “Eroom” back to Moore?
The answer to this – as with many 21st-century challenges – lies in enhanced research through AI, machine learning and natural language processing. In this post we’ll discuss the process of drug discovery and show how new technology offers a promising way out of mounting costs and inefficiencies.
What is the actual cost of drug discovery?
Estimates vary wildly, but producing a new drug costs anywhere between $314 million and $2.8 billion. The time it takes to get a new drug to market ranges between 10 and 15 years. The clinical trials phase absorbs the bulk of this time and money, but costs mount at each stage along the way. And long before a subject reaches the clinical phase, companies must dedicate substantial time and resources to Research and Development. The median capitalized R&D cost for new drugs is around $1.1 billion. It has never been more urgent for pharmaceutical and life sciences research companies to trim costs and boost efficiency – ideally both at once. Thankfully, that’s exactly the win-win that AI can deliver.
Speed: machine Learning finds patterns in huge amounts of data, in an instant.
The number of relevant data points in the life sciences continues to balloon, outstripping humans’ ability to ingest, process and categorize huge amounts of data. Machine learning answers the need for rapid, accurate pattern recognition across millions of data points. Even more importantly, artificial intelligence of this kind doesn’t need to be explicitly programmed. Instead, it learns over time and adapts to the user’s preferences.
The use cases within drug discovery alone are numerous. Researchers can use this technology to identify correlations between certain molecules and adverse clinical outcomes. Or they can analyze thousands of research records, clinical trial results and articles in minutes to identify salient trends for directing research away from dead ends.
On target: improving accuracy in Research and Development
Minimizing human dependencies in the R&D process doesn’t just make it faster – it reduces error. Through predictive analytics, sophisticated software can predict which substances are likely to have therapeutic value based on their molecular structure or other properties. This eliminates much of the guesswork associated with traditional drug discovery methods, resulting in more accurate results that can help reduce waste during later stages of drug development. Some studies have shown that machine learning models can replace clinical trials with simulations, or eliminate the need for a trial altogether by retrieving relevant results of an existing trial that may otherwise have been missed.
A way to beat the “Throw Money At It” tendency
As Scannel noted, one of the main tactics companies have employed to accelerate drug discovery has been to ramp up hiring of R&D personnel, thereby “throwing money” at the problem to make it go away. This is risky in the best of times – and we don’t live in the best of times for high tech hiring. Amidst ongoing, chronic and global talent shortages, companies (even large ones) can not afford to solve their speed problem by exacerbating their cost problem.
If they adopt intelligent software instead, they won’t have to. AI has the potential to reduce costs associated with drug discovery processes by increasing efficiency throughout each step, from early discovery right through to the regulatory and surveillance phases. By reducing manual labor required for data analysis tasks, and leveraging predictive analytics to identify promising compounds earlier in the process, companies can realize significant savings across each stage of a given project.
Similari: at the forefront of AI-enhanced drug discovery
With the right approach to AI, the pharmaceutical industry can finally make a dent in the multi-billion dollar price tag of developing new drugs. At Similari, we believe that can be done, quickly, and without scaling costs. We have brought together expertise from the domains of AI, machine learning and NLP to create a solution that enables accurate, always-on surveillance of life sciences data, on a scale that was previously out of reach for most pharmaceutical companies.
To learn more about how Similari’s state-of-the-art platform and industry-leading capabilities could push your drug discovery process to new heights, schedule a demo with our team. We’re really excited about the future, and we’re ready to show you how we plan to get there.